A Hierarchical Optimization Approach in Hybrid UAV-Enhanced MEC Networks Integrating Three-Sided Matching and Potential Game Model

Abstract

This paper investigates the joint optimization of UAV deployment and computation task offloading in a hybrid UAV-enhanced mobile edge computing (MEC) network comprising high-performance computing UAVs (H-UAVs) and relay UAVs (R-UAVs). In this network, all UAVs serve as access nodes (ANs) and computing nodes (CNs) for mobile terrestrial devices (MTDs), while R-UAVs additionally function as relay ANs that can forward offloaded tasks from their covered MTDs to adjacent H-UAVs. To tackle the combinatorial complexity, we propose a hierarchical framework integrating two core innovations: a three-sided matching game with hybrid pReferences for task offloading among MTDs, ANs, and CNs, and a state-based potential game that embeds the offloading state into UAV deployment optimization. The inner-layer matching game achieves stable matching via individually rational hybrid preferences, while the outer-layer potential game converges to a stable-state equilibrium (SSE) through marginal-contribution-based payoffs. We prove the existence of the SSE and develop a joint distributed iterative algorithm to attain it. Numerical results demonstrate the superiority of the proposed approach over baseline methods.